摘要Exclusion has been applied as a main measure for re-vegetation all over the world.This paper,by comparing the results of year-round exclusion,seasonal exclusion,and non-exclusion,quantified the vegetation variations under three different exclusion measures and their correlation to soil factors.The analysis results for community species component and plant diversity using multi-response permutation procedures (MRPPs)showed that exclusion did change the species component and increase plant diversity remarkably,while the period of exclusion had no significant influence on these two community features.The indicator species analysis and calculation of similarity indices indicated that community for year-round exclusion were becoming xerophytization and unpalatability,and showed highly spatial heterogeneity of plant species distribution,whereas community for seasonal exclusion was under stable non-equilibrium condition.Detrended correspondence analysis (DCA)and detrend canonical correspondence analysis (DCCA)results of relationship between plant species and soil variables demonstrated that soil moisture was a controlling factor for plant species component,microbiotic soil crust cover,soil organic matter,and soil bulk density had significant effects on soil moisture,among which microbiotic soil crust was a leading factor owing to its limitation to rainfall infiltration on the one hand,and its constraints to entrance of herbaceous seeds into soil or to germination of soil seeds on the other hand.As a result of long-term removal of animal grazing,crust kept intact in year-round exclusion community,which was a main reason of community xerophytization.It was also obvious from ordination results that some important environmental factors,such as tempo-spatial change of rainfall and corresponding tempo-spatial change of soil moisture,were neglected during direct gradient analysis.In addition,biodiversity was close related to soil nutrients as well as to soil moisture condition (soil water content and crust cover),and it had positive relation to available N,and negative relation to available P.Higher soil N had advantage to non-leguminous plants growth on nutrition-poor sand land definitely.The impact of P to community component was unclear and should be studied from plant physiology.Further researches on nonequilibrium theory in semi-arid rangeland will provide a scientific and flexible animal development paradigm for being implementing livestock fen-raising and grazing-forbidden policies in China.